Mar 2024 – Jun 2024 · 3 mos · Bengaluru, Karnataka, India · On-site

• I have worked on Incruiter AI interview platform, Incbot, which assesses candidates' communication skills and confidence during his/her interviews. For communication skills, fine-tuned a large language model named Hubert using supervised audio data. For confidence, developed a CNN model with MTCNN for face segmentation and also tested GPT-4o vision model for this task
• I have also worked on various Prompt Engineering tasks, including generating final reports for candidates post-interview, creating job descriptions, generating interview questions, calculating candidate answer scores, and more

Sep 2023 – Dec 2023 · 3 mos · Noida, Uttar Pradesh, India · On-site

• I have designed a prompt that takes an initial poorly formulated problem statement as input and uses GPT-3.5 turbo model to generate a refined and concise problem statement based on the original one and deployed it using Flask.
• I worked on improving practice problem editorials on the GeeksforGeeks website by identifying questions lacking editorial content. Using Prompt Engineering, I generated detailed AI editorials for better understanding and solutions, and deployed them using Flask

Mar 2023 – Aug 2023 · 5 mos · Gurugram, Haryana, India · On-site

• During my internship, I had the opportunity to work on two exciting projects. The first project was centered around Locational Intelligence, and it involved developing a smart AI-based watch for children. The main objective of this watch was to enhance child safety by incorporating geofence technology. By leveraging this technology, parents could easily track their children's whereabouts and ensure their safety within specific designated zones.
• In the second project, I was involved in Data Engineering tasks. The project dealt with a company's data, which was stored in a compressed format (zip) on an AWS S3 Bucket. The goal was to extract this data, transform it as needed, and load it into a Timescale Database for further analysis and insights. To achieve this, I utilized the Apache Nifi tool, which is an Extract, Transform, Load (ETL) tool capable of efficiently handling data flow and integration processes. With Apache Nifi, I designed data pipelines that automated the extraction, transformation, and loading of data from the S3 Bucket to the Timescale Database. This approach streamlined the data management process, making it more efficient and reliable.